Detection of maize tassels for UAV remote sensing image with an improved YOLOX Model

Abstract

Maize tassels detection is essential for future agronomic management in maize planting and breeding, with application in yield estimation, growth monitoring, intelligent picking, disease detection, etc. Nevertheless, some problems are gradually becoming more prominent for it. Maize tassels shown in the field are widespread occlusions and differ in size and the morphological color of different growth stages. Aiming at these issues, this study proposes the SEYOLOX-tiny model that detects maize tassels in the field more accurately and with robustness. Firstly, the data acquisition method better balanced the image quality and image acquisition efficiency and obtained maize tassels images from different periods to enrich our dataset by unmanned aerial vehicle (UAV). Moreover, the robust detection network extends YOLOX by embedding an attention mechanism to realize the extraction of critical features and suppressing the noise caused by adverse factors (occlusions, overlaps, etc.), which could be more suitable and robust for operating in a complex natural environment. Experimental results verify the current work hypothesis and show that the mean average precision (mAP@0.5) was 95.0%. The mAP@0.5, mAP@0.5-0.95, mAP@0.5-0.95(area=small), and mAP@0.5-0.95(area=medium) average increased by 1.5, 1.8, 5.3, and 1.7%, respectively than the original model, the proposed method which can meet the precision and robustness of vision system in maize tassels detection.